Ste in . E d w a r t I (1996) : Without domi Keasnn. N e w YV o g e l . Jonathan | 1 9 9 9 | : ' I he N e w Re le \an t A l te t i i a t i ve ' i I h e o r y " . In James H

ToMihe ihn inl.i. I'hiloM'phital rtiw/ui lives. IJ: l:pi.\icniol

706 Suitliis I'sillos

N i i n i l i i o l o (119991) ami Psi l los (| I999|) , A l though these books cliffer vastly in

their approaches and in their subslanlive theses, they can al l be seen as

participating in a c o m m o n project: that o l characterising carefully the main

features o f the realism debate and offering new ways of either exp lo r ing o ld

arguments oi thinking in novel terms about the tiebale itself. Il wouldn"! be an

exaggeration to say that the real ism debate is currently going through its

renaissance period.

In what l o l l o w s I try to appraise the present state o f the sc ien l i l i c real ism

debate with an eye to important but hitherto unexplored suggestions and open

issues that need l u i l l i e i work. In Sect ion 2. I mostly focus on the relationship

between .scientilic real ism and truth. In Sec l ion 3, 1 discuss the grounds for

realists ' epistemic op t imism.

2 What is scientific realism?

2,1 Modesty and presuinptuousness 1 offer the f o l l o win g three theses as constitutive o f scient i l ic real ism. Each o f

these is meant to dist inguish realism from a particular non-realist approach.

T h e mc laphy . s i ca l thesis: 1 he wor ld has a delinite and mind-ini lependeni

structure.

T h e .semaii l ic thesis: S c i e n l i l i c theories should be taken at lace value. They

are t ru lh-condi l ioned descriptions o f their intended domain , both observable

and unobservablc. Hence, they are capable o f being true or false. The

theoretical terms featuring in theories have putative factual reference. So , i f

sc ient i l ic theories are Hue, the unob.servable cnl i t ies they posit populate the

wor ld .

T h e ep i s t emic thesis: Mature and p i ed i c l i ve ly successful scientific theories

are we l l con l i rme i l and approximately hue of ihe wor ld . So, the enl i l ies posited

by them, or, at any rate, entities very s imi lar to those posi led, inhabit Ihe wor ld .

The /i/-.s7 thesis/);(';

78 Sidlhis I'sillos

possibly be knowable? Two options then appear to be open to would-be realists: either to coniproinise the presumptuous claim or else to compromise the modest claim.

2.2 Compromising presumptiiousncss Mere the cue is taken from Popper's (| 1982]). Take realism lo be a thesis about the aim of science (truth), leaving entirely open the issue of whether this aim is (or can ever be) achieved. Implicit in this strand is the idea that truth is understood realist-style (in the sense of correspondence with the world), in order not to compromise the modest claim as well. Popper is famous for claiming both that we can never say that this aim has been achieved, and that truth is somehow magically approached by the increasing verisimilitude of successive theories; 'magically' because there is nothing in Popper's account of verisimilitude which, even if it worked,' guarantees that there is a 'march on truth'. Musgrave (| 1996], p. 23) agrees that realism is (along with a semantic thesis) an axiologiciil thesis: '.science aims for true theories.' There is clear motivation for this compromise: even if all the theories we ever come up with are false, realism isn't threatened (ibid., p. 21). Musgrave doesn't think that all our theories have been or will be outright fal.se. But he does take this issue (whatever its outcome may be) lo have no bearing on whether realism is a correct attitude to science.^ There are, however, inevitable philosophical worries about the axiological characterization of realism. First, it seems rather vacuous. Realism is rendered immune to any serious criticism stemming from the empirical claim that the science we all love has a poor record in truth-tracking (see Laudan |I984|). Second, aiming at a goal (truth) whose achievability by the scientific method is left unspecified makes its supposed regulative role totally niyslerious. Tinally, all the excitement of the realist claim that science engages in a cognitive activity that pushes back the frontiers of ignorance and error is lost.

So the thought .seems irresistible that if realists lake seriously the claim that they must compromise, then their only real option is to compromise their modesty, not their presumptuousncss: if the world isn't in any heavyweight way independent of us, then its knowability can be safeguarded. The modest claim can easily be compromised, by coupling realism with an epistemic notion of truth which guarantees that the truth (and hence what there is) does not lie outside our cognitive scope.

' Its problems have been shown by Mi l ler (I I974|) am) Tichy {| 19741)

' It 's a bit ironic thai Van Fraassen (| i 9801, p. 8) also characlcriws realism as an axiological thesis together with a (non-l 'op |K-rian) ilo.xaslic altitude, i ; . - that acceptance of a theory involves belief in its truth.

The Present Stale oj the Scientific Realism Debate 709

2.3 Compromising modesty Here the main cue is taken fioiii Putnam's (| 19811). Take realisin to involve an epistemic concept of truth, that is a concept of truth which guarantees that there cannot be a divergence between what an ideal science will assert of the world and what happens (or what there is) in the world. This line has been exploited by Ellis (11985)) and Jardine (|I986)). For Ellis, truth is 'what we .should believe, if our knowledge were perfected, if it were based on total evidence, was internally coherent and was theoretically integrated in the best possible way' (11985], p. 68). There are many problems with this view that I won't rehearse here (see Newton-Smith | l989bl ; Psillos |I999]. pp. 2.5.1-55). The only thing I will note is that it's not at all obvious whether the suggested theory of truth can deliver the goods. To use Jardine's ([19861. p. 35) words, the ret|uited concept of truth should be neither too '.secular', nor too 'theological'. It should avoid an awkwaid dependence of truth on the vagaries of our evolving epistemic values, whilst linking truth to some notion of ideal epis-temic justification. But in the attempt to break away from 'secular' notions of truth and lo make truth a standing and stable property, it moves towards a 'theological' notion: the justification procedures become so ideal that they lose any intended connection with humanly realizable conditions. In the end, the concept of truth becomes either 'secular', resulting in an implausible relati-vism, or else 'theological', and hence not so radically different from a (realist) iion-cpislemic understanding of truth, according to which truth outrmxs the possibility of (even ideal-limit) justification. To be sure. Putnam (119901. p. viii) has dissociated his views on truth from the (Peircean) ideal-limit theory on the grounds that the latter is 'fantastic (or Utopian)'. Still, his proposed alternative, which ties ascriptions of truth to the exemplification of 'sufficicniiy good epistemic situations' fares no better than the Peircean (hcoiy vis-a-vis the sccular/lhcological test above. For one can always ask: what other than the truth (realist-style) of a proposition can guarantee that the suKicieiitly good conditions of justification obtain?^

Wright has an interesting dual thesis i) that a suitable epistemic concept of liiith docs not necessarily compromise the modesty of scientific realism and ii) thai Ihc best hope for the presumptuous claim of scientific letilism rests on a broadly verilicationist (epistemic) understanding of truth. F'or Wright, scientific realism stands mainly for a) anti-ieductionism and b) the claim that theoretical discourse is apt for 'representation or fit with objective worldly states of affairs' (| 1992), p. 159). So the first pari of his dual thesis (| 19921, PP-158-59) stems from the thought that the anti-reductionist stance of semantic realism, which treats theoretical discourse as apt for representation, is consistent with a (suitably) 'evidentially constrained' account of truth. This

' hot a llcsh look at I'ulnam's Ij l>)SI || internal realism', sec Ni ini luolo j I C'h. 7).

710 Sfiilhi.s Psillo.s

is so because, he c l a i m s , sc ienl ihc realists may accept hoib that theoretical

assertions fa i thful ly represent w o r k l l y stales-of-alfairs ami that these slatcs-of-

affairs are 'in principle' cleleclable (and hence, in pr inc iple verif iable) . In

part icidar. these w o r l d l y stales-ol ' -allairs need not be repiesentabic in a

humanly i n l c l l i g i h i c way. Hence the w o r l d ends up beinj ; i inlependcnt of

human representation (as scientif ic real ism requires), yet the w o r l d ' s in p r i n -

c iple detectability suggests a notion of truth w h i c h is suitably ' c v i d c n l i a l l y

constrained", and hence epistemic. (The motto for W r i g h t ' s ver i l i ca t io i i i s i

sc ienl ihc realists w o u l d be: theie is no in pr inc iple undetectable truth.) The

second part of W r i g h t ' s dual thesis stems from (he thought that the realists"

epistemic o p i i n i i s m requires that 'the harvest o f best methods is ( l ike ly to be)

t i u l i i anil may, quo so harvested, be reasonably so regarded (j l')86|. p. 262).

Hut. he goes on . i f Irtith is not taken to be what is ' essc i i l ia l ly certif iable by best

method" (as a veri l icat ionist lealist w o u l d have it), then there is no guarantee

that truth is achievable . So. Wright c l a i m s , the door is left open either to a van

f-raassiaii sceptic, o re l se to a Quinean pragmatis l w h o 'cashes out" talk of truth

in terms of talk of a(n) (everlasting) set o f s i m p l i c i t y - g u i d e d adjustments in our

e v o l v i n g network of beliefs in response to e m p i r i c a l a n o m a l i e s . '

In order to cast some light on (and to cr i t ic ize) W r i g h t ' s thesis, I note the

f o l l o w i n g . In his ground-breaking ((I992|), Wr ight presents a ' m i n i m a l i s t '

concept of truth (not to be confused wi th I lorvvich's (| I99()|) account) w h i c h is

characterized by some 'syntactic and semantic platitudes" (e.g. r a i s k i " s T -

schema. good behaviour with respect to negation, a ' thin" correspondence

intui t ion , stability cl al.). These platitudes (on this proposal) guarantee that a

certain discourse with a truth-predicate in it is asseitoric (apt lor truth and

fals i ty) , hut leave open the i|ueslion whether the concept of truth has a more

robust substance. N o w . some lealists bel ieve that the concept of truth has this

more robust (non-epis lemic) substance w h i c h is caplinetl by a thick" notion of

corresporKlence with reality, viz. that the soiare of the tEUth of theoretical

asseitions is w o r l d l y stales-of-alfairs . Ihis notion i s taken by lealists to be

epis temical ly unconstrainetl . Wright Juxtaposes this realist notion ol ' truth with

an epis temical ly constrained one: ' superasscrt ibi l i ty ' (| l

712 Stcilhis Psillos

i nvo lved in the defence o f rea l ism, it should be a correspondence account, in

order to safeguard the c l a i m that the wor ld is independent in its existence and

nature from what we believe. H o r w i c h ( |1997 | ) , on the other hand, after

declar ing that the sc icn i i l i c real ism debate is about the independence and

accessibi l i ty of facts about unobservable entities, takes the v iew that a ' de l l i i -

t ionary ' conception of truth (which is itself l ightweight and metaphysical ly

neutral) is all that is needed for the defence o f scient i l ic l e a l i sm . Hi s core

thought is that the t iuth-piedicate doesn't stand for any complex property, but

is a quasi - logical device for forming geneial isal ions over pioposi t ions . One

can of course pit Devi t t ' s defence of correspondence truth against M o r w i c h ' s

def la l ionism. But the serious phi losophica l issue that nonetheless remains is

H o r w i c h ' s (| I997 | ) thesis that the scienti l ic real ism debate can be fully stated

and explained without any substantive (i.e. non deflationary) concept o f truth.

In par t icula i , H o r w i c h (//(>/., p. 36) c la ims that, even when the concept o f truth

is exp l i c i t l y used in a realist (or anti-realist) thesis, e.g. when lea l is ls say that

.science achieves theoretical truth, or when instrumentalists say that theoretical

hypotheses are truth-valueless, or when vcrificationists say that all truths are

verifiable, even then i l can be captuied by a deflationist understanding of truth.

But I doubt that this is so easi ly established. W h e n realists say, for instance,

that theoretical d iscoiuse should be understood l i tera l ly , they imply that

theoretical assertions shouldn' t be taken as tianslatable into a vocabulary

that is commit ted only to observable statcs-of-affairs. The notion of translat-

abi l i ty (or its lack) may inevi tably involve reference to sameness (difference)

o f truth-conditions, wh ich , arguably, are not pait o f the resources avai lable to

the deflationist (see f-ield 11992|, pp. 324-2. ' i ) . A n important issue at stake,

w h i c h is (|uile independent o f the scient i l ic real ism debate, is whether a

deflationist story is explanator i ly complete as an account of truth, viz. whether

it expla ins everything that theie is to know about the concept of truth in a

discourse. Dissent ing voices (e.g. Tield | I992 | ; I'apineau j 19931) have made

the point that a deflationist account cannot ade(|uately expla in successful

act ion, w h i c h at least in tu i t ive ly , is taken to requite that the iniih-coiiiliiioiis

of the beliefs that guide this action have been realised. In any case, this is a key

area in w h i c h further research and argimient are necessary."^

2.4 Conceptual iiidepciulence and cpisteniic luck

Despite a l l these attempts lo force a compromise on scientific realists, neither

modesty nor presiunptuousncss has to go. Il doesn't fo l low from the c l a i m that

'' Jennings (| 1^X91) has tried lo .uliciilale 'scicniilic ( |nasi-icnlisnr in Ihc spiril o\n (I l')S4|). T l ic main llioiiplil is llial i |nasi-icalisls can 'earn llic rijilil' lo lalli ahoiil llic Irnlli cir-lalsily ol l l icolics. willioul Ihc concoinilani connililrnerHs lo a realisi onlolojzv: Ihc jiosiled Ihcorclical cnlilies inhahil a 'proicclcd' world, Whci l ie i lllis is a genuine middle way has liccn challenged hy I'iric (| l ')X(ih| | and Musgrave (| 199()|).

The Present Stale of the Scientific Realism Debate 713

the wor ld is independent of human cogni t ive activity that either human

inquirers are cogni t ive ly closed to this wor ld or else when Ihcy come to

know i l , they somehow constitute it as the object o f their investigation. Fine

(I I986b l , p. 151) has tried to make a meal out o f this purported impl ica t ion o f

the mind-independence c l a i m . But I think he is mistaken. It al l depends on how

exactly we understand the realist c l a im o f mind-independence. It should be

taken to as.seit the logical-conceptual independence of the wor ld , in the sense

that there is no conceptual or logical l ink between the truth of a statement and

our abi l i ty to recognise it, assert it, superassert it or the l ike . The entities that

science studies and finds truths about are deemed independent of us (or of mind

in general) not in any causal sense, but only in a logical sense; they are not the

outcome (whatever that means) of our conceptualisations and theorizing. A l l

this is consistent wi th the c l a im that science and its methodology are causally

dependent on the wor ld . Indeed this causal contact wi th the wor ld is presup-

posed by the lea l i s l s ' c l a i m that our methods of interaction wi th the wor ld can

be such that, at least in favourable circumstances, they can lead to the forma-

tion o f warranted beliefs about the 'deep structure' of the wor ld . Despite

several pages of phi losophical argumentation that this contact with the

independent wor ld is impossible because it wou ld amount to 'getting out o f

our s k i n ' (see Rorty [ I 9 9 l | . pp. 46fO, or because it 's 'conceptual ly

contaminated ' (Fine 11986b|, p. 1.51), i t 's a s imple truth that our ( inevitably)

conceptual give-and-take with the wor ld need not lead to the neo-idealist (or

neo-Kanl ian) thought that the causal structure of the wor ld is a lef lcct ion (or

projection) o f our concepts and theories. The independence of the wor ld

needn't be compromised . A n d it cannot be compromised unless one adopts

the implausible v iew that wor ld ly entities aie causally constituted as entities by

our conceptual and epistemic capacities and practices.^' T o be sure, realists

must grant that their 'epistemic o p t i m i s m ' that science has succeeded in

t racking truth presupposes a certain epistemic luck: i t 's not a priori true that

science has been or has to be successful in truth-tracking. If science does

succeed in truth-tracking, then this is a radically contingent fact about the way

the wor ld is and the way scientific method and theories have managed to ' latch

onto ' it (see B o y d 119811). So the presumptuousness of realism is a contingent

thesis that needs to (and can) be supported and explained by argument that

shows that the an ip l i a t ive -abduc t ive methods o f science can produce

theoretical truths about the wor ld and del iver theoretical knowledge.

If neither modesty nor presumptuousness need compromis ing , isn't there

sti l l an issue as to how presumptuous scient i l ic realism should be? I think we

should reflect a little on what exactly the phUosophical problem is. I take it to

be the fo l lowing : is there any strong reason to believe that science cannot

'' l-or Hie deleiicc of all Ihis, .see my (| 1999|, pp. 24.'i-4f>) and Ni ini luolo (| 19991, |)p. 94-9,'i).

714 Sitillii.y I'silliis

i ichieve theorelical Irull i? Tha i is. is (here any reason to beheve that, after we

l iave mitlerstoocl the theoretical stalcinenls of scient i l ic theories as expressing!

genuine propositions, we can never be warianlecl in c l a i m i n g that they are true

(or at least, more l ike ly to be true than false), where truth is iniclerstoocl realist-

style? 'I'liere are .some subtle issues here (to which we shall return below), but

once we see the problem in this way, it seems obvious that what realism should

i i np ly by its prcsumptuousness is not the implaus ib le thought that we ph i lo -

sophers should decide w h i c h scient i l ic assertions we should accept. W e should

leave that to our best science. Rather, presiiniptuousness impl ies that theore-

t ical truth is achievable (and knowable) no less than is observational truth. T h i s

c l a i m , however, may have a thin and a iliuk version. The thin version is

defended by L e p l i n ( | I 997 | ) . H i s ' M i n i m a l h!pistemic R e a l i s m ' is the thesis

that 'there are possible empi r ica l condit ions that wou ld warrant attributing

some measure o f truth to theoriesnot merely to their observable conse-

quences, but to theories themselves ' (ibid.. 102). A s L e p l i n is aware (p. 121),

many realists opt for a thicker version. This version should imply (and be

engaged in the defence of the c la im) that the a m p l i a l i v e - a b d u c l i v e methods of

science arc reliable anil can confer j u s l i h c a l i o n on theoretical asseitioiis. This

thick vers ion. I think, is the enduring result o f B o y d ' s contr ibut ion to the

defence of realism.^ Hut why do we need it? W e need it because a ' th in ' account

cannot a l low for rational or wananted belief in the iniobservable entities jiosileil

by science (and the asseilions made about them). The thin c la im merely asserts a

counterfactnal connection between .some possible empirical conditions ami the

truth o f some theorelical assertions. I h i s cannot be the litmus test for .scientilic

realism because, suitably uiidersto(xl, i t 's universally acknowledged that Ihis

connection is possible. Not only does the thin c la im fail to guarantee that Ihis

possible connection may lie actual (a condit ion icquired for the belief in the tiuth

o f A Iheoielical assertion), but any attempt to give such a guarantee would have to

engage the reliability of the mctluxl wh ich connects some empi i i ca l condit ion

with the truth of a theoretical assertion. Hence, the defence of the rationality and

reliabil i ty o f these methcKls cannot be eschewed.

T o me this means that the presumptuous strand in realist thought should be

thick. But others may disagree. He ie there is del ini te ly space for more discus-

s ion. One issue that needs to be explored as hinted at by Suppc (j 1989). pp.

3 4 0 - 6 ) and developed by Ki tcher (j I99.1a|. ("h. ."!)is how standard episte-molog ica l theories o f j u s l l l i c a l i o n . re l iabi l i ty and bel ief fo inia t ion can be

employed in the real ism debate. It may lu in out, as I ( | I 9 9 9 | , pp. 8.1-86)

and .Suppe (j 1989), p. ?>f>2) bel ieve it w i l l , that the debate on scient i l ic realisnt

is best contlucted in the context o f broader externalist ep is temologica l theories

about the nature of knowledge , jus t i l i ca t ion , etc.

' l-or :i ilcfeiKC, see 1 riuil (| Ch. .1) ;iiul my (I l 'W| . Cli 4).

The rresent Stole of the Scientijic Realism Dclnite 715

.So far, we have resisted the c l a im that the concept of truth implicated in

scienti l ic realism should be something less than a 'correspondence with

real i ty ' . The relevant pressures have led some realists to back down (e.g.

("licfc | I 9 8 8 | , p. 6). Others, however, have tried to explicate the notion o f

correspondence in such a way as to remove from it any sense in which it is

'metaphysical ly myster ious ' . O f these attempts. K i t c h c r ' s (j I99.1b|, pp. 1 6 7 -

69) stands out because he shows that Ihis notion a) need not comini t us to an

implausible view that we should (or need to) compare our assertions wi th the

wor ld and b) arises out of the idea that a lit between representations and reality

explains patterns o f successful action and intervention. A correspondence

account of truth is just another theory that can be judged for (and accepted

on the basis of) its explanatory merits.

3 EpLsteniic optiniisin I t s hard lo exaggerate the role that Sel lars played in the realist turn dur ing the

1960s. H i s attack on the 'myth o f the g iven ' and his endorsement o f the

' sc ient i l ic image", according to which what is real is what successful scienti l ic

theories posit, priori t ised scienti l ic theories over folk theories of the 'manifest

image ' as our guide to what there is (see Church land ( ( I979 | ) . .Smart (j I96.^|,

p. .^9) and M a x w e l l ( | I 9 6 2 | , p. 18) fo l lowed suit by offering arguments for

real ism based on the explanation of the success o f .science. If al l these

unobservable entities don ' t exist, i f theorelical assertions aie not we l l - con-

l i rmcd and true descriptions of an unobservable wor ld , then it isn't possible to

expla in the empi r ica l success o f science and the observation o f predicted

correlations among observable entities. Putnam ( l l*)??) , p. 7.1) turned all

this into a famous slogan; realism ' is the only phi losophy of science that

docs not make the success of science a mi rac le ' . Hence, the we l l -known 'no

mirac le ' aigument lor real ism, l o be sure, the central thought in this argument

is that realist assertions offer not the only but the best explanation of the

success of science. Be that as it may, the point of the argument is that the

success of scient i l ic theories lends credence to the fo l lowing two the.ses: a) that

scient i l ic theories should be interpreted real is t ical ly and b) that, so interpreted,

these theories are wel l conlirrned because they entail we l l - con l i rmed predic-

tions. The or ig inal authors of the argument, however, didn ' t piri an extra stress

on novel predictions, wh ich are, presumably, the li tmus lest for the ab i l i ly o f

al ternalivc approaches lo science lo expla in the success o f science. For it is

on ly on a realist understanding that novel predictions about the phenomena

come as no surprise. Yet there has been notorious disagreement as to how

exactly the conditiorrs of novelty should be understood. A novel predict ion has

been taken lo be the prediction o f a phenomenon whose existence is a.scer-

la incd only alter a theory has predicted it. This, however, cannot be the whole

716 Slothis P.sillits

story since theories also pet support from expla in ing already known

phenomena. .So i f s been suggested (e.g. W o r r a l l | I 9 8 5 | ) that the ' temporal

view" o f novelty should be replaced by a 'novelty-in-use" v iew: a pretliction of

an already known phenomenon can be use-novel with respect to some theory

provided that information about l l i i s phenomenon was not iisccl in the con-

struction of the theory. Yet il"s been notoriously d i f l i c u l l to make precise the

intuit ive idea o f 'use novelty". A fresh analysis comes f iom L c p l i n (| l ' )97 | . p

77) who analyses 'novelty" by reference to two condit ions: ' independence ' anti

'uniqueness". The thrust is that a predict ion ot a phenomenon () is l u n e l foi a

theory T i f no infoinia t ion alioul () is necessary lor the piedict iot i of () by 1',

and i l at the l ime T explains and predicts (). no other theory 'provides any

viable reason to expect" (). If these rei |uircmcnts are salisliecl. it 's hard to see

what other than the relevant truth of the theory T could best expla in the novel

predictions."

W h y . then, has the realist turn come uniler so much pressure'.' The main

target o f the non-realist onslaught has been realism"s epis tcmic op t imi sm. Note

that the or ig inal . S m a r t - M a x w e l l formulat ion of the no miracle" argument

rested on the assumption that once semantic real ism is eslablished. bel ief in the

truth o f genuinely successful scientific theories is (almost inevi tably) rat ionally

compe l l i ng . V a n I'laassen's (| I98()|) reaction to this was that the a b d u c t i v e -

a m p l i a l i v c methodology o f science fails to connect robustly empi r ica l success

and truth: two or more mutual ly incompat ib le theories can nonetheless be

empi r i ca l ly congruent and hence equal ly empi r i ca l ly successful. G i v e n that al

most one o f them can be true, semantic real ism can be retained but be

acconqianied by a sceptical attitude towards the truth o f scienti l ie theories.

N o w . realists face a d i l emma . A s Newton -Smi th (| I978 | . p. 88) pointed out.

realists can c l i n g on an 'Ignorance Response" or an 'Ar rogance Response". O n

the first horn, realists choose to hang on to a realist metaphysics of an

independent wor ld , but sacrifice their epistemic op t imi sm. O n the second

horn, they try to secure some epistemic op t imism, but sacrifice the

independence o f the wor ld by endorsing a v i ew that denies that there are

' inaccessible facts" w h i c h can make one o f the many r iva l theories true. In a

way . van I j a a s s e i r s own attitude amounts to the 'Ignorance Response" . ' 'As for

the 'Ar rogance Response". il"s hard to see how one can be a realist and sti l l

* I w c i roviowcrs (it l .cplin's hook (SnikMi | l'WX|-. l . M d y i i K i n 11 W l | ) i i if i ic lhal Ihc I ' n i i | i i c M i c s ^ CoiKhliori (I l(') is loo sli img: given IIKII S O U R - olht-i Ihcorv'I' nuiy prodicl () soon alloi I I K I S l i r s i p u H J i c l c i l i l . w l i y shouhl we iicci-|il lh:il Ihc ( M c c i d c i i h i l I h is lo i ic : i l prcccilciicc o i l ' m : i k c s () novel lor Thi l l mil I'orT'? In hi irncss lo Ixpl i i i . il i s c r i i c i : i l lh;il 1' s i i l i s l i c Ihc Inilcpcnilcncc ('onililion (IC). II n o l . t" is nol :i conlcni lc i . II )cs, () c a n he saiil h ) he prima lacic iio\eI lor hi*th X and '1 Then, nalurally. Ihc laihiie ol I K ' caiinol make O s h i l l Ihc cpislcniic halancc in l a v i u i i ol cilher 1

o r T ' and more evidence shonid he soiighl aller.

V a n l-raasscn challenges Ihc leali.sls' cpislcniic opl imism. Unl . unlike Wrigh l ami like scientilie realists, he lakes scientilie stalcillenis 'lo have liuth-coiulilitins cntilelv indcpcmlcntiv ol luinian activity Ol knowledge' l | l ' ) l l | . p. (X)

The Present Slate of the Scientific Realism Debate 717

endorse it. F'or. lo say the least, ' t r inmi ing d o w n ' the content of the wor ld so

that i l contains no inaccessible facts leaves three options available (all of wh ich

should be repugnant to realists). First, lo reinterpret the empi r ica l ly equivalent

theories so that tliey are not understood literally and the apparent conflict

among them doesn't even arise (an option taken by some L o g i c a l Empir ic i s t s ) .

Second, to adopt an epistemic notion o f truth that makes i l the case that only

one o f the empi r ica l ly c(|uivalent theories passes the truth-test (see Jardine

11986|). Third, to argue that all these theories are true, thereby l e l a t iv iz ing the

concept of truth (sec some l ime-s l ice of (Juine j I97.'i | . pp. .127-28).

C a n realists eschew the 'Ignorance Response".' The gist of van F'raassen's

challenge is that the explanatory virtues that arc part and parcel o f the

abduc t ive -ampl ia t ive methodology of science need not (and perhaps

cannot) be taken to be truth-tropic. Hence, any realist hope of forgoing the

' Ignorance Response ' by grounding their epistemic op t imism on explanatory

considerations seems to vanish. Not so fast, though. Putnam's enduring con-

tribution to the realist cause is his thought that the defence of realism cannot be

a piece o f a priori epistemology, but must rather be part and parcel o f an

empir ical-natural is t ic programme which c la ims that real ism is the best e m p i r i -

cal hypothesis o f the success of science (| I978 | . p. 19). Cap i t a l i z ing on this

thought. B o y d ( | I 9 8 I | . 11984)) embarked on an attempt lo establish the

accessibi l i ty of (and rationality o f bel ief in) theoretical truth by trying lo

defend the re l iabi l i ty o f abductive- ampliat ive inferences. Th i s w e l l - k n o w n

ahdiHtivc defence of realism starts from the fact that the heavi ly theory-laden

scientific methodology is instrumentally reliable (i.e. it y ie lds correct predic-

tions and is empi r i ca l ly successful) and argues that the best explanation of this

instrumental re l iabi l i ty is that the background theories (which inform and

dictate Ihc methods used by scientists) are approximately true. Th i s is a

phi losophical (second order) Inference lo the Best Explanat ion ( I B E ) wh ich

suggests that there is a contingeni (a posteriori) l ink between abduc t ive -

anipl iat ive methodology (and the concomitant notion of 'best explanat ion ' )

and truth. If successful, this argunient grounds realists" epistemic op t imism. It

also removes the s l ing from the r iva l argument from the underdeterminalion o f

theories by evidence ( I ITE) . I'or two empi r i ca l ly equivalent theories may not

(as a matter o f contingent fact) be equal ly good in their explanatory virtues.

Hence one o f them may wel l be the best explanation o f the evidence and

command rational be l ief . ' "

In any case. L I T E rests on two questionable premises. (I): for any theory T

there is at least another one incompal ib le theory 'I'' wh ich is empi r i ca l ly

congruent wi th T . " (II): i f two theories arc empi r ica l ly equivalent, then they

"' Tor a proper lorimilatioii and delcnee ol tliis argnnicnt. sec my l | l ' W ) | . pp. 78-81) . " I or a delcnee o l (I), see Ki ikla l | l 'WX| . C h . .S).

718 Slcithis I'sillos

are epis te inica l ly cq i i iva len i loo (that is. equal ly conl irniet l or supported by the evidence) . Both premises have been forceful ly chal lenged by realists. Some (e.g. N c w i o n - S n i i l h 11')87|) have chal lenged (I) on the grounds that the thesis it encapsulates is not p i o v e n . Note , in passing, lhal realists should be happy with a local scept ic ism. It may turn out that some domains o f inquiry {e.g. the deep structure o f space-time) are beyond our ken. Others (e.g. C d y m o u r |I98()|; H o y d 11981 j ; l .audan and L c p l i n j I W l j ; Laudan 1 1 9 % | ) have objected to (II). I le ie there are. on the face o( it. two strategies avai lable . One (Ma) is to argue that even i f we take only e m p i r i c a l evidence in the strictest sense of the w o r d as bearing on the epis lemic support of the theory, it does not f o l l o w lhal the class of the observational consequences ol the theoiy is co -ex icns ional w i t h the class o f empir i ca l facts that can lend support lo the theory. A n obvious counter-example to the c l a i m of co-ex le i i s ional i ty is that a theory can get indirect support f rom evidence it does not direct ly entai l . Ihc other strategy ( l ib ) is lo note lhal theoretical virtues are epis lemic in character and hence can bear on the support of the theory, l l e i e again there are two options avai lable to realists: ( l i b . I) to argue (rather i m p l a u s i b l y in my view) lhal some theoretical virtues are constitutive marks of truth (e.g. M c M u l l i n |I9H7|): or ( l lb .2 ) to argue for a broad concepl io i i of evidence w h i c h lakes the theoretical virtues to be broadly e m p i r i c a l and contingent marks of truth (see B o y d 1 1 9 8 1 1 : C h u i c h l a n d |I98.S|; L y c a n j 1988]). ( l lb .2 ) is an a l l rac l ive strategy for two reasons: a) it challenges the strictly empir ic is t concept ion of evidence and its l e la l io i i to rational bel ief ; b) it removes, i f successful , the apparent tension between modesty and pre-sumpiuousncss , without also forging an a prior i l ink belween theoretical virtues and truth, ( l lb .2 ) is perhaps the most d i f f i cu l t posi t ion to defend, but on closer inspection it may w e l l turn out lhal ( l la ) and ( l lb .2 ) are, at root, the very .same strategy (.see Ps i l los j 1999], pp. 1 7 1 - 7 6 ) . ' '

Not a l l defenders of rea l i sm lake the abduc l ive defence o f I B l i to be central in the defence o f real ism. There are a few specif ic problems here and one broadly phi losophica l problem. The specific problems regard the notion o f explanat ion and the condit ions under w h i c h it can be cal led b e s f . Some leal is is countenance speci l ic forms o f causal explanat ion (e.g. S a l m o n I 19841 lor the so-cal led ' c o n u n o n cause pr inc ip le " , or C a r l w r i g h t 1198.^| for ' inference to the most probable c a u s e ) but deny that these can generalise lo engender a blanket notion o f IBI-.. Others (e.g. L i p l o n | 1991 j) try to provide (descr ipt ively) an account of when a (potenlial) explanat ion is best and then lo tell a story as to when this explanat ion l icences in le ience . In the same boat, N i i n i l u o t o (|I999|, pp. 18.'i-92) sketches a formal model o f IBI- in w h i c h the 'best explanat ion ' is l i n k e d to the 'best c o n l i r m e d ' hypothesis , g iven the evidence, l - ina l ly , there are those (e.g. M i l l e r |I987|) who argue that there

l i ' i : i i i i l i i i i i c 111 s d i n c . soc i i i l c o i i M r u c l i v i M s ' i t N i i i i o i i In U l 1;. see l . i n K l a n (| I WCi|. pp. .S(l 5.1).

The Present Stale of the Scientific Realism Debate 719

cannot be a general mode of inference cal led I B E , but instead that local ampliat ive inferences in science are l icensed only when they are backed up by ' topic -spec i l i c t ru isms ' , thai is pr inciples w h i c h are so entrenched that no-one in the speci l ic domain can seriously deny them. T h i s last posit ion, how-ever, is sensitive to the issue of what renders these principles ' truisms' i f not the fact that they have been arrived at by a legit imate applicat ion of I B E .

What I cal led the 'phi losophica l p r o b l e m ' of the abductive defence o f real ism has caused a heated di.scussion. It has been wide ly argued (see L a u d a n |I984|, p. 134; van Fraassen j 19851, P- 2.55; Fine |1986a|, 11986b]) that the realists' use o f (a second-order) I B E in defence of realism is circular and (|ueslion-begging. For , the thought is, it takes for granted the rel iabi l i ty of a mode o f inference w h i c h is doubted by non-realists. T h i s challenge has led some realists to question the v iabi l i ty of the abductive strategy. N e w t o n - S m i t h (|1989a|, p. 179) for instance has cal led the realism associated with this strategy ' f a d e d ' . A n d Harre (|19881) left behind 'truth rea l i sm' and its 'deeply flawed" abductive defence in favour of a methodological strategy that he cal led ' p o l i c y r ea l i sm' (see also Hendry 11995)) .

T h i s issue is the focal point of the debate at present. A proper appreciation of what is at stake presupposes a better understanding of the broader episteino-log ica l agendas of the participants. A s is expla ined in detail in my (| 1999], C h . 4), the abductive defence of real ism can only proceed wi th in a broad natur-al ist ic f ramework in w h i c h the charge of c i rcular i ty loses its bite, because what is sought is not jus t ihca l ion of inferential methods and practices (at least in the neo-Cartesian internalist sen.se) but their explanat ion and defence (in the epis temologica l externalist sense). It's not as if the abductive defence o f real ism should persuade a commit ted opponent o f real ism to change sides. Strict empir ic is ts , for instance, are not l i k e l y to be m o v e d by any defence of I B E , be it c i rcular or straight, precisely because as M c M u l l i n (| 1994|, p. 100) has noted. Ihey s i m p l y choose to tolerate unexplained regularities and phe-nomena. (One such regularity is that science has been inslrumental ly reliable and successful.) V a n Fraassen's insistence that the explanatory virtues are merely pragmatic is just a further twist to this tolerance of the unexplained. So, strict empir ic is ts reject the abductive defence of real ism not so much because i t ' s c i rcular (they w o u l d reject a defence o f I B E even i f it were straight), but m a i n l y because they refrain from accepting the existence o f unobservable entities on any grounds that transcend those w h i c h can be derived from naked-eye observations (see S a l m o n |I985|). But unless this attitude is itself the most reasonable to adopt (and 1 doubt whether it is), it doesn ' t f o l l o w that I B E is u n r e a s o n a b l e . " N o r does it f o l l o w that the employment of I B E in an abductive defence of the re l iabi l i ty of I B E is quest ion-begging and unpersuasive. M a n y

" V a n Fraassen (|I989|. pp 160-70) tins suggested that tlil-.-eonceived as a n i l e i s incdherenl. t iarman (| I9%1) and Doiiven (| 1W9|1 liave rcliuucd lliis c la im.

720 Sltilhi\

( i f not all) use iiuxlusponens in i re l lec t ive ly as a sound inrcrcni ia l rule and yet

an establ is l inieni o f the soundness o f modus ponens at the nieta-level by an

argument w h i c h effectively uses modus ponens can sti l l expla in to Ihcni why

and in virtue o f what features deductive reasoning is sound. In any case, realists

vary in the extent to w h i c h they adopt an abductive defence o l the re l iabi l i ty o f

I B E . There aie (hose brazen realists, l ike B o y d | ( I 9 8 I ) | , Trout (| I998 | ) and

Ps i l los ( ( l 9 9 9 ) | w h o take the charge o( c i rcular i ty seriously and try to meet it

wi th in a naturalistic perspective. One ccn t ia l thought in this camp is that

abduction is the only general mode of amplia t ive reasoning and i f this fails,

then no amplia t ive reasoning (ami hence no learning from experience) is

possible. Then there are the temperate realists (see I x p l i n | I 9 9 7 | . p. 116)

w h o stress that abduction and induction are distinct modes of reasoning and try

to argue that I B E is no worse than ordinary irrductions w h i c h are O K for non-

realists. F i n a l l y , there arc realists ( l ike B r o w n 1 1 9 9 4 | . C h . I) who side-step the

charge o f c i rcular i ty and argue that the explanatory story told by realisnt is just

more adventurous and enl ightening than alternative stories.

There is also a deep empi r ica l chal lenge to real ism and its abductive

defence: the Pessimist ic induct ion . A s l.audan ( | I 984 | ) has pointed out. the

history o f science is replete wi th (heories lirat were once cons i i le rcd to be

empi r i ca l l y successful and fruitful, but w h i c h turned out to be fal.se and were

abandoned. If the history o f scieirce is the wasteland o f aborted 'best theore-

t ical explanat ions ' o f the evidence, then it might w e l l be that current best

explanatory theories might take the route to the wasteland in due course. Not

a l l realists l ind this argunrcnt threatening. Sorire (e.g. D e v i l l | I 9 8 4 | ) (ind it

s imply (arrd. I think, correct ly) overstated. Others (e.g. A ln rcde r 1 1 9 9 2 1 ) take a

'blirrd realist ' stance: at any given stage o f ini | i r i ry some o f our theoretical

be l ie ls are trire. yet we carr never tell w h i c h these arc because 'we have no

reliable way o f determining w h i c h o f our currently complete ly authorised

beliefs w i l l suffer truth-value revis ion in the future' (p. 1 7 8 ) . " Yet others

(e.g. H a c k i n g (11984))) answer the pe.ssimislic induct ion by choos ing to be

realist about entities w h i c h can be manipulated by wel l -designed experiments

and noi about theories. This last v i e w k n o w n as entity realismis cr i t ic ised

in my ( ( I999 | ) . pp. 22.'S-8). What about those o f us who think that we shouki

take seriously the I'essirrristic Induction and try to meet it?

A l t h o u g h other strategics may be avai lable . I think the best defence o f

real ism is to try to reconci le the historical record wi th some form o f real ism.

In Older to do that, realists shouki be more selective in what they are realists

Ajwrl fidiii noliiii: i l i u l |K-isisU-iil rcu-nliiiii Ml llic lliucirelicnl level may he ;i leliahle (hiil liillihie)

way In siiiplc oiil the Iheoielicnl asscilicms thai wdn' l siilTcr Initli-valiic revision, one can claim

lhal Alri iedei 'ss lai ice InlKiiray lo ihe 'prcraee parailcix'. I l e ( | I W 3 | . I'. 1 SOI li ics K m i m i l c r lliis

c la im, l i u l a related piul i lem still remains. A 'hliiul leaiist' asseils hoili that hn each and every

theoretieal assertion I' we can't know whether it is Hue imil IWM we know that soirie I's ate Iriie.

l iven il this joint asseitioii is corisislent. Ihe liisl pail lemoves any hasis lor aceeplini; the second.

The Present Stale of the Scientific Realism Debate 721

about. A c l a im that now emerges wi th some force is that theory-change is not

as radical and discontinuous as the opponents of scientific real ism have

suggested. Realists have aimed to show that there are ways to identify which

theoretical constituents o f abandoned scientific theories essentially contributed

to their successes, separate them from others that were ' i d l e ' o r as Ki tcher

has put it. merely 'presupposit ional pos i t s 'and demonstrate that those

components w h i c h made essential contributions to the theory's empir ica l

success were those that were retained in subsequent theories t)f the same

domain (see Ki t che r | l 9 9 3 a | ; Ps i l los [1999]). If the relevant realist arguments

are sound, then Ihe fact that our current best theories may wel l be replaced by

others does not, necessarily, undermine scientific real ism. A l l it shows is that a)

we cannot get at the truth a l l at once; and b) our inferences from empir ica l

support to approximate truth should be more refined and cautious in that they

should commit us only to the theoretical constituents that do enjoy evidential

support and contribute to the empi r ica l successes of Ihe theory. Realists ground

their epistemic op t imism on the fact that newer theories incorporate many

theoretical constituents o f their superseded predecessors, especial ly those

constituents that have led to empi r ica l successes. The substantive continuity

in theory-change suggests that a rather stable network of theoretical prij iciples

and explanatory hypotheses has emerged, w h i c h has survived revolutionary

changes, and w h i c h has become part and parcel of our evo lv ing scientific

image of the wor ld .

Th i s reaction to the Pessimist ic Induction was ini t ialed by W o r r a l l ' s seminal

(119891) . What he cal led 'structural rea l i sm' is an a l l empi to capitalise on the

fact that despite the radical changes at Ihe theoretical l eve l , successor theories

have tended to retain the mathematical structure of their predecessors. Wor ra l l ' s

thought is that theories can successfully represent the structure of the wor ld ,

although they tend to be wrong in their c la ims about the entities they posit.

Despite its ini t ia l appeal, it turns out that this particular posit ion is very difficult

to defend (see L a d y m a n 11998] ; Ps i l los 11999] , C h . 7; Ps i l los | for thcoming | ) .

Car twr ight (| 1999]. p. 4) has taken a different path. She is happy to go from the

' impress ive empi r ica l successes of our best physics theories' to 'the truth of

these theories ' , but she denies that the assertions made by these theories are

universal in scope. Rather, she goes for a ' loca l real ism about a variety o f

different kinds o f knowledge in a variety o f different domains across a range of

h ighly differentiated situations' {ibid., p. 2.^) wh ich tallies wi th her view that

the wor ld is best seen as disunified, wi th no laws or principles hold ing across

the board and across different domains o f inquiry. T h i s is certainly an issue that

needs to be investigated further. Partly because it seems to be an open issue

whether real ism is inherently l inked with the aim of unification (but .see

Ki tcher | I 9 9 9 | ) . B u i main ly because, arguing as Cartwright does, for local

truths which may vary from one model to another and from one domain to

7 2 2 Slalliis I'sillo.s

another, may well invo lve a perspecl ival notion o f truth wh ich has character-istics not suitable for realisn).

A s is we l l known, realists talk of approximate Irudi and lake science and its methods to issue in (al least most typ ica l ly ) approximate ly true bel ie ls . H o w much of a substantive concession this is is a matter or dispute. Laudan (| I984|) c la ims that the realist cause is doomed unless a I'ormal semantic for approx-imate truth is in the oHing . G ie rc (| I988|) concedes this but c la ims that realists can do wel l with a notion ol s imi lar i ty between the theoretical model and the doiTiaiii to wh ich il applies. A ronson , Harre and Way (| I994|) try to make good on the notion of s imi lar i ty by dev is ing an informal account of approximate truth that rests on the v iew that theories are type-hierarchies of natural kinds. Others (e.g. N i i n i l u o l o 11999]) sti l l think that there are good prospects for a formal (and coi isistei i l ) expl icat ion of i ipproximate truth. M y own v iew (see 119991, C h . I I ) is (hat we shouldn ' t be deterred in our phi losophica l disputes by formal issues if the operative notions are intui t ive ly clear and do not lead to paradoxes. A s .Smith (| I998|) has suggested, (he in(uitive notion of ' approx-imate truth" can be expl icated suf l i c i cndy wel l to be usable along the fo l l ow ing l ines: for a stalemeni P. ' P ' is approximate ly (rue i f f approxinuKely P. Th is relega(es much (o (he coiicep( of approx imadon . but there is no reason to think that a domain-speci l ic understanding of approximat ion is not robust enough to warrant a.scription of appr()xima(e truth in statements about each doma in .

A l though, as we have seen, theie have been ext i cmely important and profound challenges to real ism, (he only articulated rival phi losoj ihical position that has emeiged is van Praassen 's Constiiiciive Ewpiiicism (| I98()|).''^ This v iew is al ieady famil iar to eve iyone and has been thoroughly debated in Church land and Hooker (I I98.'i|). Its core point is that commiKed empiricis(s cannot be forced to be scientific reali.sts because a) they can olTcr an al(erna(ivc accoun( of science which (akes science (o a im at empir ica l adeijiiacy and which involves only belief in the empir ica l adequacy of theories: and b) (his accoun( of science is complete in the sense that there are no features of science anil its pracdce which cannot be accounted for (or explained away) from this empiricist peispective. ( J iven that it is impossible to do Justice to (he massive literature on Ibis subject in the present space (but see Rosen j I994| and my 11999] , C h . 9). I shall only make a general comment about (he spiri( of van Iraasscn ' s approach . ' " A s M i l l e r ( | |987| . p. .V)9) nicely pu( i(, van Praassen 's stance is a k i n d o f inotlern 'pr inc ip le of

Ami. of course, l-irio's ( | ima | . |l9Kf,hl) ,|ii'lisl ,lisiniss.-,l nl ihi- philosophicnl .IchnH-iillcipcllicr. line's views are ciiliei/etl in deliiii in Mns);m\ (| l')S9| ;irul my (|l Ch, 1(1),

' Vnn I'KKisscn (| l')X,S|.2,

7 2 4 Si

726 Sfathis l\\iUos

H e n d r y . R. f I W . ^ i l : " R e a l i s m a m i P r o g r e s s : W h y S c i c n i i s r s S h o i t i d h e R e a l i s t s ' i n R. I ' e l l o w s (((/.), Philttstiphx aiul Tci hnohinv. C ' a n i h r i c l g e : C a i n h r i d ^ e U n i v e r s i t y P r e s s .

N o r w i c h . P. ( 1 9 9 0 ) : Tmlh. O x f o r d : H I a c k w e l l .

I l o r w i c h . P. | I 9 9 7 | : ' R e a l i s m a n d T r u t h ' . I'iKiutn Suulics in the I'hilo.sophv of ihc Scic/u

7 2 8 Sfiithis PsiUos

Suppe. F. (19891: The Semantic Cnmrriion of Tlwoiic.s nml Scientific Realism.

Chicago: Univcrs i ly of l lhriois Press.

T i chy . V. 11974|: 'On Popper's Dcf in i l ion of Ver is i in ihtudc" , Biilish Journal for the

Philoxi'phy of Science, 25. pp. l55-(i().

Trout. J . D. 11998|: Measuring the Intentional World. Oxford: Oxford Univcrs i l y Press,

van Fraassen. B . C . 119801: The Scientific Imafic Oxford: Clarendon Press,

vjin Fi.iasscn. H, C. 119S5|: ' l -n iph ic i sn i in Philosophy of Science", in P. M . Churchland

nnd C . A. Hooker {ah) \\'m\.

van Frnassen. U. C . 119X91: hius amiSvmmeirv. Ox i o rd : Clarendon Press. van l-raassert. B. C. 119941: 'Against Transcendenlal l - j i ipir ic isrn". in T. J . Staplelon

(cr/.). The Question of lienneneutics. Dordrecht: K luwer . van I'rnasscn. H. C. (2(KM)1: " l l i e F'alse 1 lopes of had i l i ona l Fpis le inology" . Thilosophv

and Vhenf>mentd(ii>ical Research. 611. pp. 2,^.1- 80.

W o m i l l . J . (lyS.SI: 'Sc i cn i i l i c ni.scovety and TheoryConl i r rna l ion ' . in J . PitI ((